Repeatability and reproducibility of diffusional variance decomposition (DIVIDE) imaging

Motivation: Novel Multidimensional diffusion encoding (MDE) technique, diffusional variance decomposition (DIVIDE), may provide more detailed insights into tissue microstructure. Goal(s): To evaluate the feasibility of DIVIDE imaging for human brain. Approach: Ten healthy-subjects underwent MDE (2D-EPI sequence with 29 linear and 26 spherical b-tensors) twice using 3T-MRI. Regional values of 20 ROIs was extracted for 10 DIVIDE metrics. Coefficient of variation (CV) and interclass correlation coefficient (ICC) were calculated. Results: Intra-subject CV was less than 5% in almost all regional metrics. Intra-subject CV was lower than that of inter-subject CV in all regional metrics. ICC showed almost perfect agreements for almost all regional metrics. Impact: Recently developed MDE technique, diffusional variance decomposition (DIVIDE), may be reliably used for measuring diffusion metrics with a potential to provide more detailed insights into tissue microstructure in complex tissues, such as crossing or kissing fiber configurations in the brain.